Application Of Partial Discharge Ultrasonic Signal Detection And Wavelet Packet Transform In Signal Denoising

Among power equipment failure, insulation failure has had a high proportion. PD can measure the insulation condition of equipment and is also the main factor causing the deterioration of equipment insulation. Therefore detecting and analysising the PD signal, which can discover the insulation defects existing in the electric power equipment, can avoid the occurrence of electrical accidents. This paper analyzes the characteristics of ultrasonic signals caused by PD and sets up a experimental platform of ultrasonic signal detection. For collecting PD signal of electrical equipment, use wavelet packet transform to deal with narrowband interference signal and white noise coming from ultrasonic signal about partial discharge. The main researching contents of this article are as follows:(1) A typical equivalent circuit of single-gap insulation medium is analyzed and the single-gap discharge circuit is simulated by simulink module of MATLAB software. The simulation results obtain that there are the discharging circuit in the insulating medium inside, when partial discharge occurs. The charging and discharge circuit occur voltage and current with vibration attenuation characteristics, and their frequencies are the natural frequency of the circuit.(2) The structure modules about ultrasonic detection of PD are selected and designed and the detection experimental platform of PD ultrasonic signal is built. The reasonable arrangement position is studied when the sensor detects the PD and ultrasound signal about PD coming from electric power equipment on-site is collected.(3) Using six different denoise methods including FFT transform, wavelet transform and wavelet packet transform method combining noise reduction effect evaluation index including energy ratio, similarity and SNR denoise the noisbloc signal.The analysis obtains that wavelet packet transform in the aspect of noise reduction is most satisfactory and the noise reduction index of wavelet packet:energy noise ratio is 0.9879 and similarity value is 1.1716 and SNR value is 16.0726db.(4)The adaptive wavelet packet decomposition method of de-noising is designed for removing the signal PD periodic narrowband interference and the results show that the adaptive wavelet packet decomposition de-noising method is superior to the conventional adaptive filter denoising. The denoised signal exhibits good smoothness.The adjust the threshold method of wavelet packet is designed for removing white noise of PD and the results show that this method is better than the default threshold method on denoising effect. When the threshold resizes to 9.5, it achieves better to remove the white noise. Then the noise reduction index is energy value of ratio 0.9807, the similarity value 0.7262, SNR value 15.7449db respectively.